Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
APPLICATION OF KEYNESIAN THEORY AND NEW ECONOMIC GEOGRAPHY IN PORTUGAL. AN ALTERNATIVE ANALYSIS
1. APPLICATION OF KEYNESIAN THEORY AND NEW ECONOMIC
GEOGRAPHY IN PORTUGAL. AN ALTERNATIVE ANALYSIS
Vitor João Pereira Domingues Martinho
Unidade de I&D do Instituto Politécnico de Viseu
Av. Cor. José Maria Vale de Andrade
Campus Politécnico
3504 - 510 Viseu
(PORTUGAL)
e-mail: vdmartinho@esav.ipv.pt
ABSTRACT
This work aims to test the Verdoorn Law, with the alternative specifications of (1)Kaldor (1966), for five
Portuguese regions (NUTS II) from 1986 to 1994. It is intended to test, even in this work, the alternative
interpretation of (2)Rowthorn (1975) of the Verdoorn's Law for the same regions and periods. The results of this
work will be complemented with estimates of these relationships to other sectors of the economy than the
industry (agriculture and services sectors) and for the total economy of each region. This work aims, yet, to
study the Portuguese regional agglomeration process, using the linear form the New Economic Geography
models that emphasize the importance of spatial factors (distance, costs of transport and communication) in
explaining of the concentration of economic activity in certain locations. In a theoretical context, it is intended,
also, to explain the complementarily of clustering models, associated with the New Economic Geography, and
polarization associated with the Keynesian tradition, describing the mechanisms by which these processes are
based.
Keywords: agglomeration; polarization; Portuguese regions; linear models.
1. INTRODUCTION
With this work it is intended to explain the complementarily, similarities and differences between the
clustering models, associated with the New Economic Geography (3)(Fujita et al., 2000), and polarization,
associated with the Keynesian tradition (4)(Targetti et al., 1989). It is pretended also studying the Portuguese
regional agglomeration and polarization processes. For the agglomeration process we use the linear form of the
New Economic Geography models that emphasize the importance of factors in explaining the spatial
concentration of economic activity in certain locations. The polarization process is mainly based in the very
known Verdoorn law. (5)Verdoorn (1949) was the first author to reveal the importance of the positive
relationship between the growth of labor productivity and output growth, arguing that the causality is from output
to productivity, thus assuming that labor productivity is endogenous. An important finding of the empirical
relationship is the elasticity of labor productivity with respect to output that according to Verdoorn is
approximately 0.45 on average, external limits between 0.41 and 0.57. This author also found that the
relationship between productivity growth and output growth reflects a kind of production technology and the
existence of increasing returns to scale, which contradicts the hypothesis of neoclassical constant returns to
scale, or decreasing, and absolute convergence Regional.
2. DESCRIPTION OF THE MODELS
The models of the keynesian theory and new economic geography are developed in several works like (6-
7)Martinho (2011a and 2011b).
3. DATA ANALYSIS
Considering the variables on the models referred previously and the availability of statistical
information, we used the following data disaggregated at regional level. Annual data for the periods 1986 to
1994 and 1987 to 1994 corresponding to the five regions of mainland Portugal (NUTS II), for the different
economic sectors and the total economy of these regions. These data were obtained from Eurostat (Eurostat
Regio of Statistics 2000).
4. EMPIRICAL EVIDENCE OF THE VERDOORN'S LAW
The results in Table 1, obtained in the estimations carried out with the equations of Verdoorn, Kaldor
and Rowthorn for each of the sectors of the economy and for the total economy of each of the five regions
1
2. considered, to state the following.
The industry is the sector that has the biggest increasing returns to scale, followed by agriculture and
service sector. Services without the public sector present values for the income scale unacceptable and
manufacturing presents surprisingly very low values, reflecting a more intensive use of labor.
It should be noted, finally, for this set of results the following: Verdoorn's equation is the most
satisfactory in terms of statistical significance of the coefficient obtained and the degree of explanation in the
various estimations. There is, therefore, that productivity is endogenous and generated by the growth of regional
and sectoral output.
Table 1: Analysis of economies of scale through the equation Verdoorn, Kaldor and Rowthorn, for each of the
economic sectors and the five NUTS II of Portugal, for the period 1986 to 1994
Agriculture
2
Constant Coefficient DW R G.L. E.E. (1/(1-b))
Verdoorn 0.042* 0.878*
pi a bqi
1.696 0.805 38
(5.925) (12.527)
Kaldor -0.042* 0.123**
ei c dqi
1.696 0.075 38
(-5.925) (1.750)
8.197
Rowthorn1 -0.010 -0.621**
1.568 0.087 38
pi 1 1ei (-0.616) (-1.904)
Rowthorn2 -0.010 0.379
1.568 0.034 38
qi 2 2 ei (-0.616) (1.160)
Industry
2
Constant Coefficient DW R G.L. E.E. (1/(1-b))
-12.725* 0.992*
Verdoorn 2.001 0.587 37
(-4.222) (8.299)
12.725* 0.008
Kaldor 2.001 0.869 37
(4.222) (0.064)
125.000
15.346* -0.449*
Rowthorn1 1.889 0.326 37
(9.052) (-3.214)
15.346* 0.551*
Rowthorn2 1.889 0.776 37
(9.052) (3.940)
Manufactured Industry
2
Constant Coefficient DW R G.L. E.E. (1/(1-b))
8.296* 0.319*
Verdoorn 1.679 0.139 37
(4.306) (2.240)
-8.296* 0.681*
Kaldor 1.679 0.887 37
(-4.306) (4.777)
1.468
12.522* -0.240*
Rowthorn1 1.842 0.269 37
(12.537) (-2.834)
12.522* 0.760*
Rowthorn2 1.842 0.891 37
(12.537) (8.993)
Services
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.045* 0.802*
Verdoorn 1.728 0.506 38
(-3.253) (6.239)
0.045* 0.198
Kaldor 1.728 0.059 38
(3.253) (1.544)
5.051
0.071* -0.694*
Rowthorn1 1.817 0.255 38
(4.728) (-3.607)
0.071* 0.306
Rowthorn2 1.817 0.063 38
(4.728) (1.592)
Services (without public sector)
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
-0.074* 1.020*
Verdoorn 1.786 0.609 38
(-4.250) (7.695)
0.074* -0.020
Kaldor 1.786 0.001 38
(4.250) (-0.149)
---
0.076* -0.903*
Rowthorn1 1.847 0.371 38
(4.350) (-4.736)
0.076* 0.097
Rowthorn2 1.847 0.007 38
(4.350) (0.509)
All Sectors
Constant Coefficient DW R2 G.L. E.E. (1/(1-b))
2
3. -0.020* 0.907*
Verdoorn 1.595 0.648 38
(-2.090) (8.367)
0.020* 0.093
Kaldor 1.595 0.019 38
(2.090) (0.856)
10.753
0.056* -0.648*
Rowthorn1 2.336 0.255 32
(6.043) (-2.670)
0.056* 0.352
Rowthorn2 2.336 0.225 32
(6.043) (1.453)
Note: * Coefficient statistically significant at 5%, ** Coefficient statistically significant at 10%, GL, Degrees of freedom; EE,
Economies of scale.
5. EQUATION LINEARIZED AND REDUCED OF THE REAL WAGES, WITH THE VARIABLES
INDEPENDENT NATIONALLY AGGREGATED
Thus, the equation of real wages that will be estimated in its linear form, will be a function of the
following explanatory variables:
ln rt f 0 f1 ln Y pt f 2 ln Trpt f 3 ln G pt f 4 ln pt f 5 ln w pt f 6 ln T prt f 7 ln Prt , (1)
where:
- rt is the real wage in region r (5 regions) for each of the manufacturing industries (9 industries);
- Ypt is the gross value added of each of the manufacturing industries at the national level;
- Gpt is the price index at the national level;
- is the number of workers in each industry, at national level;
pt
- Wpt is the nominal wage for each of the industries at the national level;
- Trpt is the flow of goods from each of the regions to Portugal;
- Tprt is the flow of goods to each of the regions from Portugal;
- Prt is the regional productivity for each industry;
- p indicates Portugal and r refers to each of the regions.
The results obtained in the estimations of this equation are shown in Tables 2 and 3.
Table 2: Estimation of the equation of real wages with the independent variables aggregated at national level
(without productivity), 1987-1994
ln rt f 0 f1 ln Y pt f 2 ln Trpt f 3 ln G pt f 4 ln pt f 5 ln w pt f 6 ln T prt
ln pt
Variable lnY pt lnTrpt lnGpt lnwpt lnTprt
2
Coefficient f1 f2 * f3 * f4 f5 * f6 * R DW
LSDV
Coefficients -0.038 0.674 -0.967 0.025 0.937 -0.594
0.810 1.516
T-stat. (-0.970) (4.227) (-7.509) (0.511) (15.239) (-3.787)
L. signif. (0.333) (0.000) (0.000) (0.610) (0.000) (0.000)
Degrees of freedom 290
Number of obervations 302
Standard deviation 0.146 T.HAUSMAN - 416.930
(*) Coefficient statistically significant at 5%.
Table 3: Estimation of the equation of real wages with the independent variables aggregated at national level
(with productivity), 1987-1994
ln rt f 0 f1 ln Y pt f 2 ln Trpt f 3 ln G pt f 4 ln pt f 5 ln w pt f 6 ln T prt f 7 ln Prt
ln pt
Variable lnY pt lnTrpt lnGpt lnwpt lnTprt lnPrt
2
Coefficient f1 * f2 * f3 * f4 * f5 * f6 * f7 * R DW
LSDV
Coefficients -0.259 0.557 -0.884 0.256 0.883 -0.493 0.258
0.858 1.560
T-stat. (-7.064) (4.422) (-9.671) (5.919) (19.180) (-3.996) (10.443)
L. signif. (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Degrees of freedom 289
Number of obervations 302
Standard deviation 0.126 T.HAUSMAN - 7086.989*
(*) Coefficient statistically significant at 5%.
3
4. This equation 1 estimated of real wages presents satisfactory results in terms of statistical significance
of coefficients, the degree of adjustment and autocorrelation of errors. For the signs of the estimated coefficients
that represent the respective elasticities, taking into account the expected by the economic theory, we confirm
that, apart the gross value added, the price index and the nominal wages per employee, all coefficients have the
expected signs.
6. LINEARIZED AND REDUCED EQUATION OF REAL WAGES, WITH THE VARIABLES
INDEPENDENT REGIONALLY DISAGGREGATED
Following the equation of real wages reduced and in a linear form, but now with the independent
variables disaggregated at regional level, in other words, considered only for the region being analyzed, and not
for the whole of Portugal, as in the previous equation. Although this equation does not consider the effect of
nearby regions of r in this region, aims to be a simulation to determine the effect of the regions in their real
wages, that is:
ln rt f 0 f1 ln Yrt f 2 ln Trpt f 3 ln Grt f 4 ln rt f 5 ln wrt f 6 ln T prt (2)
where:
- rt is the real wage in the region r, for each of the manufacturing industries;
- Yrt is the gross value added of each of the manufacturing industries at the regional level;
- Grt is the price index at the regional level;
- is the number of workers in each industry, at regional level;
rt
- Wrt is the nominal wage per employee in each of the manufacturing industries at regional level;
- Trpt is the flow of goods from each region to Portugal;
- Tprt is the flow of goods to each of the regions from Portugal.
Table 4 presents the results of estimating equation 2 where the independent variables are
disaggregated at regional level. About the signs of the coefficients, it appears that these are the expected, given
the theory, the same can not be said of the variable rt (number of employees). However, it is not surprising
given the economic characteristics of regions like the Norte (many employees and low wages) and Alentejo (few
employees and high salaries), two atypical cases precisely for opposite reasons. Analyzing the results in Tables
2, 3 and 4 we confirm the greater explanatory power of the variables when considered in aggregate at the
national level.
Table 4: Estimation of the equation of real wages with the independent variables disaggregated at the regional
level
ln rt f 0 f1 ln Yrt f 2 ln Trpt f 3 ln Grt f 4 ln rt f 5 ln wrt f 6 ln T prt ,
ln rt
Const. lnY rt lnTrpt lnGrt lnwrt lnTprt
Variables
2
Coefficients f0 * f1 * f2 * f3 * f4 * f5 * f6 * R DW
Random effects
Coefficients 0.101 0.629 -0.571 -0.151 0.516 -0.506 0.670 1.858
1.530
T-stat. (4.147) (4.625) (-10.218) (-5.364) (13.357) (-3.985)
(3.355)
L. signif. (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
(0.001)
0.756 1.934
0.098* 0.559* -0.624* -0.155* 0.619* -0.411*
LSDV (4.129) (4.449) (-11.380) (-6.130) (16.784) (-3.511)
Degrees of freedom 295 - 289
Number of obervations 302 - 302
Standard deviation 0.155 - 0.165 T.HAUSMAN - 72.843*
(*) Coefficient statistically significant at 5%.
7. ALTERNATIVE EQUATIONS TO THE EQUATIONS 1 AND 2
We also made two alternative estimates in order to test the existence of multicollinearity among the
explanatory variables, considering all the variables by the weight of the work in every industry and every region
in the total industry in this region for the equation 1 and the weight of work in every industry and every region in
the national total of this industry for the equation 2, following procedures of (8)Hanson (1998). It is noted that
the results are very similar to those previously presented to the estimates of equations 1 and 2, which allows us
to verify the absence of statistics infractions.
4
5. 8. EQUATION OF THE AGGLOMERATION
In the analysis of the Portuguese regional agglomeration process, using models of New Economic
Geography in the linear form, we pretend to identify whether there are between Portuguese regions, or not,
forces of concentration of economic activity and population in one or a few regions (centripetal forces). These
forces of attraction to this theory, are the differences that arise in real wages, since locations with higher real
wages, have better conditions to begin the process of agglomeration. Therefore, it pretends to analyze the
factors that originate convergence or divergence in real wages between Portuguese regions. Thus, given the
characteristics of these regions will be used as the dependent variable, the ratio of real wages in each region
and the region's leading real wages in this case (Lisboa e Vale do Tejo), following procedures of (9)Armstrong
(1995) and (10)Dewhurst and Mutis-Gaitan (1995). So, which contribute to the increase in this ratio is a force
that works against clutter (centrifugal force) and vice versa.
Thus:
ln rt a0 a1 ln Ynt a2 ln Trl a3 ln Lnt a4 ln Prt a5 ln RLrmt a6 ln RLrgt a7 RLrkt a8 ln RLrnt
(3)
lt
where:
- Ynt is the national gross value added of each of the manufacturing industries considered in the database used;
- Trl is the flow of goods from each region to Lisboa e Vale do Tejo, representing the transportation costs;
- Lnt is the number of employees in manufacturing at the national level;
- Prt is the regional productivity (ratio of regional gross value added in manufacturing and the regional number
of employees employed in this activity);
- RLrmt is the ratio between the total number of employees in regional manufacturing and the regional number
of employees, in each manufacturing (agglomeration forces represent inter-industry, at regional level);
- RLrgt is the ratio between the number of regional employees in each manufacturing and regional total in all
activities (represent agglomeration forces intra-industry, at regional level);
- RLrkt is the ratio between the number of regional employees in each manufacturing, and regional area
(representing forces of agglomeration related to the size of the region);
- RLrnt is the ratio between the number of regional employees, in each of the manufacturing industries, and the
national total in each industry (agglomeration forces represent inter-regions in each of the manufacturing
industries considered).
The index r (1,..., 5) represents the respective region, t is the time period (8 years), n the entire national
territory, k the area (km2), l the region Lisboa e Vale do Tejo, g all sectors and m manufacturing activity (9
industries).
The results of the estimations made regarding equation 3 are shown in Tables 5 and 6. Two different
estimates were made, one without the variable productivity (whose results are presented in Table 5) and one
with this variable (Table 6).
Table 5: Estimation of the agglomeration equation without the productivity
ln rt a0 a1 ln Ynt a2 ln Trl a3 ln Lnt a4 ln RLrmt a5 ln RLrgt a6 RLrkt a7 ln RLrnt
lt
Variab. Constant lnYnt lnTrl lnLnt lnRLrmt lnRLrgt lnRLrkt lnRLrnt
2
Coef. a0 a1 a2 a3 a4 a5 a6 a7 R DW
Random ef.
V.Coef. -3.991 -0.040 0.012 0.390 -0.413 -0.507 -0.228 0.368 0.253 1.474
T-stat. (-3.317) (-1.353) (1.469) (4.046) (-4.799) (-4.122) (-4.333) (4.249)
L. sign. (0.001) (0.177) (0.143) (0.000) (0.000) (0.000) (0.000) (0.000)
Degrees of freedom 293
Number of obervations 302
Standard deviation 0.126 T.HAUSMAN - 1.870
(*) Coefficient statistically significant at 5%.
(**) Coefficient statistically significant at 10%.
5
6. Table 6: Estimation of the agglomeration equation with the productivity
ln rt a0 a1 ln Ynt a2 ln Trl a3 ln Lnt a4 ln Prt a5 ln RLrmt a6 ln RLrgt a7 RLrkt a8 ln RLrnt
lt
Variab. Constant lnYnt lnTrl lnLnt lnPrt lnRLrmt lnRLrgt lnRLrkt lnRLrnt
2
Coef. a0* a1* a2* a3* a4* a5* a6* a7* a8* R DW
Random eff.
V.Coef. -3.053 -0.240 0.015 0.486 0.218 -0.266 -0.333 -0.141 0.230 0.455 1.516
T-stat. (-2.991) (-7.182) (2.026) (5.934) (8.850) (-3.494) (-3.102) (-3.067) (3.026)
L. sign. (0.003) (0.000) (0.044) (0.000) (0.000) (0.001) (0.002) (0.002) (0.003)
0.649 1.504
LSDV -0.307* -0.033* 0.330* 0.256* -0.049 0.011 -0.027 0.006
(-9.259) (-4.821) (5.701) (8.874) (- (0.169) (-0.968) (0.137)
0.972)
Degrees of freedom 292 - 285
Number of obervations 302 - 302
Standard deviation 0.116 - 0.136 T.HAUSMAN - 33.578*
(*) Coefficient statistically significant at 5%.
Comparing the values of two tables is confirmed again the importance of productivity (Prt) in explaining
the wage differences. On the other hand improves the statistical significance of coefficients and the degree of
explanation.
9. SOME FINAL CONCLUSIONS
In the estimates made for each of the economic sectors, with the Verdoorn law, it appears that the
industry is the largest that has increasing returns to scale, followed by agriculture and service sector.
With the new economic geography models, it appears that the explanatory power of the independent
variables considered, is more reasonable, when these variables are considered in their original form, in other
words, in the aggregate form for all locations with strong business with that we are considering (in the case
studied, aggregated at national level to mainland Portugal). On the other hand, given the existence of "backward
and forward" linkages and agglomeration economies, represented in the variables Rlrmt and RLrgt, we can
affirm the existence of growing scale economies in the Portuguese manufacturing industry during the period
considered. It should be noted also that different estimates were made without the productivity variable and
with this variable in order to be analyzed the importance of this variable in explaining the phenomenon of
agglomeration. It seems important to carry out this analysis, because despite the economic theory consider the
wages that can be explained by productivity, the new economic geography ignores it, at least explicitly, in their
models, for reasons already mentioned widely, particular those related to the need to make the models
tractable.
For this period the new economic geography and the Keynesian theory say the same about the
Portuguese regions development.
10. REFERENCES
1. N. Kaldor. Causes of the Slow Rate of Economics of the UK. An Inaugural Lecture. Cambridge: Cambridge
University Press, 1966.
2. R.E. Rowthorn. What Remains of Kaldor Laws? Economic Journal, 85, 10-19 (1975).
3. M. Fujita; P. Krugman; and J.A. Venables. The Spatial Economy: Cities, Regions, and International Trade.
MIT Press, Cambridge 2000.
4. F. Targetti and A.P. Thirlwall. The Essential Kaldor. Duckworth, London 1989.
5. P.J. Verdoorn. Fattori che Regolano lo Sviluppo Della Produttivita del Lavoro. L´Industria, 1, 3-10 (1949).
6. V.J.P.D. Martinho. The Verdoorn law in the Portuguese regions: a panel data analysis. MPRA Paper 32186,
University Library of Munich, Germany (2011a).
7. V.J.P.D. Martinho. Regional agglomeration in Portugal: a linear analysis. MPRA Paper 32337, University
Library of Munich, Germany (2011b).
8. G. Hanson. Regional adjustment to trade liberalization. Regional Science and Urban Economics (28), 419-
444 (1998).
6